Skip to content

Latest commit

 

History

History
62 lines (46 loc) · 1.76 KB

README_streamlit.md

File metadata and controls

62 lines (46 loc) · 1.76 KB

Burmese News Classification — Streamlit UI

This folder contains a ready-to-run Streamlit app for your Burmese news classifier.

1) Project layout

your_project/
├─ app.py
├─ models/
│  ├─ bilstm_mynews.keras
│  ├─ nb_tfidf.joblib
│  ├─ keras_tokenizer.pkl
│  ├─ label_encoder.pkl
│  ├─ config.json              # should include {"max_len": <int>} (or similar)
│  └─ stopwords.txt            # optional
└─ requirements.txt

2) Setup

Windows (Python 3.10 recommended)

# from project root
py -3.10 -m venv .venv
.venv\Scripts\activate

py -3.10 -m pip install --upgrade pip
py -3.10 -m pip install -r requirements.txt

If TensorFlow/protobuf prints version warnings, pin protobuf to 3.20.*:

py -3.10 -m pip install "protobuf==3.20.*" --force-reinstall

3) Run

python -m streamlit run app.py
# or
py -3.10 -m streamlit run app.py

Then open the local URL shown in the terminal (usually http://localhost:8501).

4) Notes

  • BiLSTM model: the app reads config.json for max_len. If missing, it defaults to 200.
  • Tokenizer: must match the one used during training (keras_tokenizer.pkl).
  • Label encoder: used to convert predicted indices back to class names.
  • Stopwords: optional. If present, they are applied during preprocessing.
  • Burmese tokenization: the app will use PyIDAUNGSU if available; otherwise it falls back to a simple regex-based tokenizer so it still works.

5) Deploy (optional: Streamlit Community Cloud)

  1. Push your repo to GitHub.
  2. On Streamlit Community Cloud, create a new app, point it to app.py.
  3. Set Python version to 3.10 in the advanced settings, and make sure your requirements.txt is in the repo.